Bias and Trend Correction of Precipitation Datasets to Force Ocean Models

Abstract

Abstract A novel method to adjust the precipitation produced by atmospheric reanalyses using observational constraints to force ocean models is described. The method allows the preservation of the qualities of the high-resolution and high-frequency output from the reanalyses while eliminating their bias and spurious trends. The method is shown to be robust to degradation in both space and time of the observation dataset. This method is applied to the ERA-Interim precipitation dataset using the Global Precipitation Climatology Project (GPCP) v2.3 as the observational reference in order to create a debiased dataset that can be used to force ocean models. The produced debiased dataset is then compared to ERA-Interim and GPCP in a suite of forced ice–ocean numerical experiments using the GFDL OM4 model. Ocean states obtained with the new precipitation dataset are consistent with results from GPCP-forced experiments with respect to global metrics but produces the extra sea surface salinity variability at the time scales unresolved by the observation-based dataset. Discrepancies between modeled and observed freshwater fluxes are discussed as well as the strategies to mitigate them and their impacts.